Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Speed up writeVInt #62345

Merged
merged 11 commits into from
Sep 15, 2020
Merged
Show file tree
Hide file tree
Changes from 9 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 0 additions & 1 deletion benchmarks/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -78,7 +78,6 @@ cd fcml*
make
cd example/hsdis
make
cp .libs/libhsdis.so.0.0.0
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This was just wrong.

sudo cp .libs/libhsdis.so.0.0.0 /usr/lib/jvm/java-14-adoptopenjdk/lib/hsdis-amd64.so
```

Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,109 @@
/*
* Licensed to Elasticsearch under one or more contributor
* license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright
* ownership. Elasticsearch licenses this file to you under
* the Apache License, Version 2.0 (the "License"); you may
* not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/

package org.elasticsearch.benchmark.search.aggregations.bucket.terms;

import org.apache.lucene.util.BytesRef;
import org.elasticsearch.Version;
import org.elasticsearch.common.bytes.BytesReference;
import org.elasticsearch.common.io.stream.BytesStreamOutput;
import org.elasticsearch.common.io.stream.DelayableWriteable;
import org.elasticsearch.common.io.stream.NamedWriteableRegistry;
import org.elasticsearch.search.DocValueFormat;
import org.elasticsearch.search.aggregations.BucketOrder;
import org.elasticsearch.search.aggregations.InternalAggregation;
import org.elasticsearch.search.aggregations.InternalAggregations;
import org.elasticsearch.search.aggregations.bucket.terms.StringTerms;
import org.openjdk.jmh.annotations.Benchmark;
import org.openjdk.jmh.annotations.BenchmarkMode;
import org.openjdk.jmh.annotations.Fork;
import org.openjdk.jmh.annotations.Measurement;
import org.openjdk.jmh.annotations.Mode;
import org.openjdk.jmh.annotations.OutputTimeUnit;
import org.openjdk.jmh.annotations.Param;
import org.openjdk.jmh.annotations.Scope;
import org.openjdk.jmh.annotations.Setup;
import org.openjdk.jmh.annotations.State;
import org.openjdk.jmh.annotations.Warmup;

import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
import java.util.concurrent.TimeUnit;

@Fork(2)
@Warmup(iterations = 10)
@Measurement(iterations = 5)
@BenchmarkMode(Mode.AverageTime)
@OutputTimeUnit(TimeUnit.MILLISECONDS)
@State(Scope.Benchmark)
public class StringTermsSerializationBenchmark {
private static final NamedWriteableRegistry REGISTRY = new NamedWriteableRegistry(
List.of(new NamedWriteableRegistry.Entry(InternalAggregation.class, StringTerms.NAME, StringTerms::new))
);
@Param(value = { "1000" })
private int buckets;

private DelayableWriteable<InternalAggregations> results;

@Setup
public void initResults() {
results = DelayableWriteable.referencing(InternalAggregations.from(List.of(newTerms(true))));
}

private StringTerms newTerms(boolean withNested) {
List<StringTerms.Bucket> resultBuckets = new ArrayList<>(buckets);
for (int i = 0; i < buckets; i++) {
InternalAggregations inner = withNested ? InternalAggregations.from(List.of(newTerms(false))) : InternalAggregations.EMPTY;
resultBuckets.add(new StringTerms.Bucket(new BytesRef("test" + i), i, inner, false, 0, DocValueFormat.RAW));
}
return new StringTerms(
"test",
BucketOrder.key(true),
BucketOrder.key(true),
buckets,
1,
null,
DocValueFormat.RAW,
buckets,
false,
100000,
resultBuckets,
0
);
}

@Benchmark
public DelayableWriteable<InternalAggregations> serialize() {
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm unsure if we actually want this benchmark, especially compared to the one that @jimczi showed me. But it is fairly targeted which can be useful.

return results.asSerialized(InternalAggregations::readFrom, REGISTRY);
}

@Benchmark
public BytesReference serializeVint() throws IOException {
try (BytesStreamOutput buffer = new BytesStreamOutput()) {
buffer.setVersion(Version.CURRENT);
for (int i = 0; i < 1000000; i++) {
buffer.writeVInt(i);
buffer.reset();
}
return buffer.bytes();
}
}
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we probably don't want to keep this benchmark but I pushed it so you could see what I was using for the numbers I shared.


}
Original file line number Diff line number Diff line change
Expand Up @@ -218,12 +218,26 @@ public void writeInt(int i) throws IOException {
* using {@link #writeInt}
*/
public void writeVInt(int i) throws IOException {
final byte[] buffer = scratch.get();
/*
* Shortcut writing single byte because it is very, very common and
* can skip grabbing the scratch buffer. This is marginally slower
* than hand unrolling the entire encoding loop but hand unrolling
* the encoding loop blows out the method size so it can't be inlined.
* In that case benchmarks of the method itself are faster but
* benchmarks of methods that use this method are slower.
* This is philosophically in line with vint in general - it biases
* twoards being simple and fast for smaller numbers.
*/
if (Integer.numberOfLeadingZeros(i) >= 25) {
writeByte((byte) i);
return;
}
byte[] buffer = scratch.get();
int index = 0;
while ((i & ~0x7F) != 0) {
do {
buffer[index++] = ((byte) ((i & 0x7f) | 0x80));
i >>>= 7;
}
} while ((i & ~0x7F) != 0);
buffer[index++] = ((byte) i);
writeBytes(buffer, 0, index);
}
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -61,7 +61,7 @@
import static org.hamcrest.Matchers.sameInstance;

/**
* Tests for {@link BytesStreamOutput} paging behaviour.
* Tests for {@link StreamOutput}.
*/
public class BytesStreamsTests extends ESTestCase {
public void testEmpty() throws Exception {
Expand Down Expand Up @@ -827,6 +827,16 @@ public void testVInt() throws IOException {
final int value = randomInt();
BytesStreamOutput output = new BytesStreamOutput();
output.writeVInt(value);

BytesStreamOutput simple = new BytesStreamOutput();
int i = value;
while ((i & ~0x7F) != 0) {
simple.writeByte(((byte) ((i & 0x7f) | 0x80)));
i >>>= 7;
}
simple.writeByte((byte) i);
assertEquals(simple.bytes().toBytesRef().toString(), output.bytes().toBytesRef().toString());

StreamInput input = output.bytes().streamInput();
assertEquals(value, input.readVInt());
}
Expand Down